stelloprint retweeted
OK, this is IMPORTANT to understand for any and all software engineers:
Inference engineering 101: reading from the KV cache is very cheap; speculative decoding makes sessions more efficient. Coding use cases make HEAVY use of both.
THIS is why inference is SUPER profitable!!
stelloprint retweeted
Open-source is essential. I built MongoDB, one of the world’s most popular open-source databases. Now I’m building Viam, an open-source harness for physical AI.
I don’t want a world where millions of developers are dependent on the paywalled technology of just a handful of companies. No one should.
I want to see 1B+ new physical AI deployments over the next ten years.
Here’s why open source is essential to getting there. 🧵
We seeded 2m canaries into popular AI services and tracked where they were stolen and abused.
This identified:
- Three compromised providers
- Resellers stealing prompts via built-in logging
- Providers ignoring ZDR & privacy flags
dragos.com/blog/ai-supply-ch…
stelloprint retweeted
Open weights are open. Anyone can download them, test them, fine-tune them, and run them on their own hardware. They’re the most inspectable kind of AI there is. A model’s country of origin doesn’t tell you whether it’s safe or good. Benchmarks and evals do.
and open source has never had a passport. Linux started in Finland, Python in the Netherlands, Ruby in Japan, Lua in Brazil, llama.cpp in Bulgaria. Kernel commits come from Intel, Google, Red Hat, Huawei and 2,000+ developers each release. Nobody puts a banner on their site about that.
Same with open models and the tools that serve them: Mistral from France, Falcon from the UAE, Qwen and DeepSeek from China, Llama and gpt-oss from the US. Roughly 33% of vLLM’s contributors are based in China, and about 52% for SGLang, so need disclosure for all of these also? Huawei was the top employer by kernel changesets in Linux 5.10 and 5.14. I can go on....
No one puts a banner on their site about those. Most technology stack is already "multinational" with open contributions for public use.
stelloprint retweeted
Akamai, a $12B company, built a SECOND version of its website that only AI agents could see.
its AI search visibility increased 364%, while citations jumped 85%.
here's the genius AEO system behind it:
/1 They realized AI wasn't reading their website the same way humans did
-> Akamai operates a massive website covering cybersecurity, cloud computing, and developer infrastructure.
-> but its content wasn't consistently appearing when people asked AI about the products it offered.
-> their team believed part of the problem was how AI agents were accessing and interpreting those pages.
/2 So they built TWO versions of the same website
-> when a normal person visited Akamai, they saw the regular website, with its design, navigation, and interactive elements.
-> but when a supported AI agent requested a page, Akamai's system delivered a different, simplified version.
-> same URL, same underlying product information, but without the extra code, menus, and other elements AI didn't need.
/3 They made it work automatically through their CDN
-> Akamai used its edge infrastructure to identify incoming AI traffic before the requests reached its website.
-> detected AI agents were routed through Scrunch's Agent Experience Platform, which delivered simpler HTML versions of Akamai's existing content.
-> unlike llms.txt (which major AI search engines don't reliably read), this approach delivers optimized content directly when supported AI agents request a page.
-> they didn't have to redesign their public website or manually rewrite every page.
/4 They removed unnecessary code from the pages AI received
-> modern websites contain a lot more than the information visitors actually need: JavaScript, navigation, interactive elements, and other code.
-> Scrunch produced machine-readable versions using Akamai's existing product pages and documentation.
-> the goal was to make the information easier for AI to retrieve and understand, without changing what human visitors saw.
/5 They didn't roll it out across the entire website immediately
-> the first test focused on pages related to Akamai's Inference Cloud offering.
-> they measured how these pages appeared across selected AI prompts before optimization.
-> then they compared pages receiving the optimized treatment against pages that weren't.
/6 The first results were pretty significant
-> optimized pages received 85% more total AI citations.
-> they were cited across 38% more unique prompts.
-> brand visibility for questions that DIDN'T mention Akamai by name increased 364%.
-> in ChatGPT, Akamai's brand presence against competitors improved 133%.
and this last point is especially important.
people weren't just asking "what is Akamai?" and getting Akamai's website back.
the company started appearing more often when buyers asked general questions about the problems its products solve.
/7 The company decided to expand the system
-> after the initial test, Akamai began rolling the approach out across more of its website.
-> it also turned the technology into a commercial product called AI Brand Presence.
-> the same infrastructure it uses to manage web traffic could now detect AI agents, change how they receive content, and measure the results.
Akamai essentially built a system where humans and AI receive different versions of the same underlying website information.
and in its initial test, that was associated with a 364% improvement in visibility for searches where people weren't even mentioning the brand.
if you want to rank #1 on AI search + turn AI traffic into pileline,
reply "GTA"
stelloprint retweeted
Excited to launch Gemini 4 Neon: frontier-level intelligence, super fast task completion, and available to Pro & Ultra users now!! We think you’ll enjoy this model. Alchemy at its finest… 🧪
it's crazy how it's always the same thing
the guy who makes being nice his personality is actually dick
anthropic made safety and calling for slowing down its personality
and meanwhile they made aggressively speeding up a big part of their IPO pitch
stelloprint retweeted
“Jensen Huang was building Anthropic's worst nightmare. Then he was ousted.”
How coincidental?
stelloprint retweeted
JUST IN: Google buys bankrupt Spirit Airlines’ massive trove of internal workplace data for $10 million to train its AI models.
stelloprint retweeted
Banning agents from reading your own country's laws is a very weird flex (looking at you, France). I mean, in what world would a country want to *hide* their legal texts? What is the purpose of preventing people from easily exploring and understanding it?
stelloprint retweeted
Health care pricing makes zero sense to me.
Case in point: My daughter just had an MRI on her foot and ankle.
The hospital billed $8,278. Insurance covers 85% of it. We owe $1,250.
A full body MRI through Function is under $900 cash. A single body part from an independent imaging center is only $300.
Our share after insurance is 4x the cash price.
At this point, it make sense to cash pay for everything.
stelloprint retweeted
Mwen pa t janm atann yon bagay konsa.
Mwen tande nou byen fò, byen klè.
Mwen renmen nou tou.
N ap triyonfe. Bondye ap ban nou laviktwa.
stelloprint retweeted
Uh oh... Jensen Huang just jumped ship, abandoning OpenAI and Anthropic; telling every nation they need their own sovereign AI.
Is it over for Sam Altman? $NVDA
stelloprint retweeted
Dario Amodei on open source models:
“If the path continues, I think we could get to a very dangerous place.”
“When you control a model and you're deploying it, you have the ability to monitor its usage.”
“You can revoke a user's access. You can change what the model is willing to do.”
“When a model is released in an uncontrolled manner, there's no ability to do that. It's entirely out of your hands.”
“We should confront the advocates of this with that problem and challenge them to solve it.”
He clearly doesn't believe Intelligence should be in the hand of the masses and Sam Altman also saying "RAM" shouldn't be accessible to everyone indefinitely agreed with him.




































